We provide a self-contained discussion of modern econometric methods useful in analyzing applied industrial-organization problems such as determining firms' market power, the effect of factors underlying this power, and the strategies used by firms. We develop and summarize each method in enough detail that a reader can apply these methods without having to search through other texts.
We concentrate on information-theoretic (IT) methods. These methods are well-suited for problems in which we lack information about the exact objective function and the data generation process, we have relatively few observations or the data are excessively aggregated, and we want to impose constraints from economic theory. If we have few observations and we do not know the underlying data generation process, the traditional maximum-likelihood (ML) approach may be inappropriate. Within the IT methods, we describe in greatest detail the maximum entropy (ME) and generalized maximum entropy (GME) methods, which are used in Chapters 9 and 10.
We start by introducing the concept of entropy and discuss several estimation techniques base on maximizing entropy. Finally, we turn to three other closely related models: empirical likelihood (EL), generalized method of moments (GMM), and the Bayesian method of moments (BMOM). Though we mostly use the GME method to analyze problems discussed in the text, the other IT methods presented in this appendix are suitable for analyzing many of these problems.